Study On High Resolution Time-frequency Analysis Of Seismic Signals And Its Application

As a powerful tool for the analysis of time-varying non-stationary signals,time-frequency analysis has become a hotspot in modern signal processing.Based on the theory of time-frequency analysis,this thesis introduces the basic theory of time-frequency analysis and several traditional time-frequency analysis methods,including STFT,CWT and WVD.Compared with the traditional time-frequency method,an adaptive time-frequency analysis method has a significant advantage in the analysis of non-stationary signals.However,In order to solve the problem of modal aliasing problems in the core algorithm EMD.We have proposed several improved algorithms that include EEMD and CEEMD,Although the modal aliasing problem has been suppressed,but in reality there are more or less modal aliasing phenomenon,and this algorithm will be the original simple signal complexity,and isolated too many IMF components.This leads to an excess of the original effective signal,which may cause the signal to be partially distorted,or to reduce the signal-to-noise ratio of the signal.The empirical wavelet transform algorithm is a new method of time-frequency analysis.The algorithm breaks the limitation of the traditional time-frequency analysis algorithm in the adaptive field.Combining the advantages of the empirical mode decomposition and the traditional wavelet transform,the complex signal can be decomposed into a more physical mode.In this thesis,the principle of the empirical wavelet transform algorithm is studied in detail,and the existing problems in the algorithm are improved and optimized.The value of the empirical wavelet transform algorithm as a new time-frequency analysis method in practical application is verified.In this thesis,we propose an empirical wavelet transform algorithm based on morpho transform,which is based on the advantages of mathematical morphology in image processing,in order to solve the problem of spectrum partitioning when dealing with complex spectrum signals.Because our seismic signals are non-stationary and complex signals,we have made higher demands on the adaptive segmentation capability of the signal and the ability to find "meaningful" modalities in the spectrum.Therefore,the Otsu method based on histogram segmentation is studied and applied tothe segmentation of signal spectrum in empirical wavelet.Finally,adaptive wavelet transform is obtained.It is a high-resolution time-frequency analysis method to apply the adaptive empirical wavelet transform to the geological forward model and the actual seismic data.Since the instantaneous properties obtained by empirical wavelet transform are more accurate and effective,the resolution and credibility are greatly improved.